1. Field of the Invention
The invention relates to a method for the production of a bottom resist for a two-layer O.sub.2 /Reactive Ion Etching system.
2. Description of Related Art
In the photolithographic production of resist structures with light having a wavelength of approximately 250 nm, two-layer O.sub.2 /RIE systems (RIE=Reactive Ion Etching) are one of the few possibilities of producing dimensionally accurate structures, even with low definition imaging lenses, when using high wafer topography. In this connection, a top resist which contains silicon, or which can be silylated, is applied to a planarizing and absorbent bottom resist. This top resist is then exposed and developed, as well as silylated, if necessary, or chemically expanded, and subsequently, the structure obtained in this way is transferred into the bottom resist by means of an oxygen plasma (see, for example: "Microelectr. Eng.," Vol. 11 (1990), pages 531 to 534).
The bottom resist must possess the following properties:
high etch resistance against substrate etching PA1 heat shape retention above 100.degree. C. PA1 non-solubility in solvents for the top resist, developer solutions and, if necessary, silylation or expansion solutions PA1 high absorption at the exposure wavelength good planarization properties PA1 easy removability by means of non-hazardous stripping media.
The latter requirement is particularly significant if the two-layer O.sub.2 /RIE system is to be used for so-called lift-off processes (see in this regard: W. Scot Ruska, "Microelectronic Processing," McGraw-Hill Book Company, 1987, pages 227 to 230).
Generally, standard Novolak resists are used as the bottom resist; these are heated at temperatures above 200.degree. C. and become insoluble in solvents and developer solutions in this manner (see in this regard: "Proc. SPIE," Vol. 1262 (1990), pages 528 to 537). In many cases, polyimide pre-stages tempered at approximately 400.degree. C. are also used as the bottom resist; at this temperature, these are converted into insoluble polyimides see: "J. Electrochem. Soc.," Vol. 135 (1988), pages 2896 to 2899, as well as "Solid State Technology," June 1987, pages 83 to 89). Such resist systems demonstrate a high proportion of aromatics and--because of the heating process--are even more stable against etching than normal photoresists based on Novolak, in which the heating process generally leads to cross-linking. However, these resist systems are so insoluble that they can no longer be removed by conventional stripping media.
Heated Novolaks can only be removed by oxidizing, strong acids, such as Caro acid, i.e. peroxomonosulfuric acid (H.sub.2 SO.sub.5), or fuming nitric acid. However, not all substrates tolerate such treatment. For example, aluminum, which is used as a material for tracks, is soluble in Caro acid and can easily be converted to aluminum oxide by oxidizing media. Stripping of the bottom resist with such media would therefore result in damage to the tracks. When using fuming nitric acid, the aluminum would be passivated, but this stripping medium is dubious from a toxicological point of view. Polyimides, on the other hand, can be removed by solutions containing hydrazine, such as mixtures of ethylene diamine and hydrazine hydrate, but there are also serious toxicological reservations against such solutions, since the maximum workplace concentration (MAK value) of hydrazine is only 5 ppm.
Frequently, resists are also removed by means of an oxygen plasma causing them to oxidize to CO, CO.sub.2 and H.sub.2 O. However in the case of two-layer systems if residues of the top resist containing silicon are still present after substrate etching, the silicon would be oxidized to SiO.sub.2, which would then remain on the wafer in the form of fine particles, and can result in defects.
To achieve good planarization, materials are also already known for bottom resists, which are cured not thermally, but photolyrically (see in this regard: "Polym. Mater. Sci. Eng.," Vol. 60 (1989), pages 385 to 389). This involves aromatic methacrylic and epoxy compounds. However, these materials demonstrate at least 20% lesser etching stability than Novolaks. In order to offset this lack of stability, thicker layers are required. The known systems furthermore require curing only by UV light. Curing can therefore already occur during plasma etching, due to the UV light that occurs in this connection.
It is also known to use an additional lift-off layer of polysulfone in two-layer O.sub.2 /RIE systems for the lift-off technique, with a Novolak layer being applied over it (see: "J. Electrochem. Soc.," Vol. 138 (1991), pages 1765 to 1769). After heating of the Novolak, the polysulfone still remains soluble (in N-methyl pyrrolidone), so that the layer structured using a resist containing silicon can be removed by means of a lift-off process. Such a procedure requires two additional layers (of different materials). However, the more additional varnishing steps are necessary, the more the defect density increases, as a rule.